Data sources
LookAlike is built entirely from freely available, open data. Every data source is keyless, has no authentication requirement, and is refreshed with pinned vintages recorded in the separate LookAlike-data-pipeline repository.
Open data sources
Section titled “Open data sources”| Source | Vintage | What it provides | Access | License |
|---|---|---|---|---|
| ACS (Census Bureau) | 2020–2024 5-year Summary File | ACS-derived demographic & socioeconomic features (income, education, age, housing, commute, poverty) and context fields aggregated to census tracts | Keyless bulk download of table-based .dat files from Census Bureau | Public domain |
| TIGER/Line (Census Bureau) | 2024 | Tract boundary polygons, centroids, and land-area measurements | Per-state shapefiles from Census Bureau | Public domain |
| Overture Maps Places | 2026-09-23.1 | US points of interest (restaurants, grocery stores, cafés, shopping) grouped into densities per km², excluding permanently-closed places and places with existence confidence below 0.5 | Public GeoParquet on S3, queried with DuckDB (no full download required; spatial predicate pushdown reads only relevant data) | CDLA-Permissive 2.0 |
| CBSA delineation (Census Bureau) | OMB 2023 | 383 Metropolitan Statistical Areas in the continental US; maps counties to metros for the metro-focus feature | Keyless .xlsx download from Census Bureau | Public domain |
| Brand store locations | 2026-10-05-01 release | For each curated brand: store coordinates from All The Places (CC0 weekly scrapes) when it covers at least 80% as many tracts as OpenStreetMap, otherwise from OpenStreetMap via the Overpass API | CC0 / ODbL | |
| Basemap | current | OpenFreeMap dark vector style built on OpenStreetMap data; no key or account | © OpenMapTiles, © OpenStreetMap contributors (ODbL) |
Data caveats
Section titled “Data caveats”Key numbers
Section titled “Key numbers”- Tracts: 83,006 continental-US census tracts in release 2026-10-05-01, across all 50 states (excluding Alaska, Hawaii, and territories); about 82,995 receive a score.
- Curated brands: 99 brands. Store counts vary by brand and source; the Brand panel names each brand’s source.
- Scored features: 18 (13 ACS-derived plus 5 Overture retail-density measures). Race and ethnicity shares are shown as context only.
- Metro areas: 383 Census-defined Metropolitan Statistical Areas for geographic focus.
Accessing the data
Section titled “Accessing the data”All raw data is downloaded and cached inside the data-pipeline repository. The pipeline writes processed assets to dist/releases/<release-id>/ and publishes them to R2:
tracts/features.f32— compact float32 feature arrays per tracttracts/sims.u8— pre-computed within-brand fit percentiles, 83,006 rows × 99 brandstracts/has.u8— brand-present flag, 83,006 rows × 99 brandstracts/centroids.f32— tract centroids (latitude, longitude)tracts/boundaries.binandtracts/boundaries.json— tract polygons for containing-tract assignment (range-served)tracts/rel.u8— low-ACS-reliability flag per tractmeta.json— brand profiles, fitted presence-model coefficients, store-source counts, feature descriptions, and coherencestores/<brand-id>.json— store coordinates for each brandmetros.json— metro-area definitions and summariestracts.pmtiles— vector tileset of tract polygons for the map; it no longer carries per-brand similarity values
The per-brand candidate JSON is generated under data/out for the offline consistency test and is not part of the web bundle; the browser rebuilds the ranked list from the binaries.
The browser loads these national assets and scores live using JavaScript. Authenticated customer uploads are stored separately from the open national release in private R2 and client-scoped D1 records.
Provenance
Section titled “Provenance”The data pipeline is maintained in its own repository. Every transformation is logged: which tables were fetched, how many rows were kept, what cells were extracted, and how features were engineered. Release manifests record the source versions, the store-source counts, and the pipeline revision.
The choice to use only free, open data is intentional. It makes the tool reproducible and auditable — no proprietary foot-traffic data, no purchased mailing lists, no commercial geospatial databases. The trade-off is that neighborhood fit on open data is a weak predictor of sales or success. See the method for how the tool validates and frames this limit.